Metron Profiler Client

This project provides a client API for accessing the profiles generated by the Metron Profiler. This includes both a Java API and Stellar API for accessing the profile data. The primary use case is to extract profile data for use during model scoring.

Stellar Client API

PROFILE_GET

The PROFILE_GET command allows you to select all of the profile measurements written. This command takes the following arguments:

REQUIRED:
    profile - The name of the profile
    entity - The name of the entity
    periods - The list of profile periods to grab.  These are ProfilePeriod objects.
OPTIONAL:
	groups_list - Optional, must correspond to the 'groupBy' list used in profile creation - List (in square brackets) of 
            groupBy values used to filter the profile. Default is the empty list, meaning groupBy was not used when 
            creating the profile.
    config_overrides - Optional - Map (in curly braces) of name:value pairs, each overriding the global config parameter
            of the same name. Default is the empty Map, meaning no overrides.

There is an older calling format where groups_list is specified as a sequence of group names, “varargs” style, instead of a List object. This format is still supported for backward compatibility, but it is deprecated, and it is disallowed if the optional config_overrides argument is used.

The periods field is (likely) the output of another Stellar function which defines the times to include.

Groups_list argument

The groups_list argument in the client must exactly correspond to the groupBy configuration in the profile definition. If groupBy was not used in the profile, groups_list must be empty in the client. If groupBy was used in the profile, then the client groups_list is not optional; it must be the same length as the groupBy list, and specify exactly one selected group value for each groupBy criterion, in the same order. For example:

If in Profile, the groupBy criteria are:  [ “DAY_OF_WEEK()”, “URL_TO_PORT()” ]
Then in PROFILE_GET, an allowed groups value would be:  [ “3”, “8080” ]
which will select only records from Tuesdays with port number 8080.

Configuration and the config_overrides argument

By default, the Profiler creates profiles with a period duration of 15 minutes. This means that data is accumulated, summarized and flushed every 15 minutes. The Client API must also have knowledge of this duration to correctly retrieve the profile data. If the Client is expecting 15 minute periods, it will not be able to read data generated by a Profiler that was configured for 1 hour periods, and will return zero results.

Similarly, all six Client configuration parameters listed in the table below must match the Profiler configuration parameter settings from the time the profile was created. The period duration and other configuration parameters from the Profiler topology are stored in local filesystem at $METRON_HOME/config/profiler.properties. The Stellar Client API can be configured correspondingly by setting the following properties in Metron’s global configuration, on local filesystem at $METRON_HOME/config/zookeeper/global.json, then uploaded to Zookeeper (at /metron/topology/global) by using zk_load_configs.sh:

    $ cd $METRON_HOME
    $ bin/zk_load_configs.sh -m PUSH -i config/zookeeper/ -z node1:2181

Any of these six Client configuration parameters may be overridden at run time using the config_overrides Map argument in PROFILE_GET. The primary use case is when historical profiles have been created with a different Profiler configuration than is currently configured, and the analyst needing to access them does not want to change the global Client configuration so as not to disrupt the work of other analysts working with current profiles.

Key Description Required Default
profiler.client.period.duration The duration of each profile period. This value should be defined along with profiler.client.period.duration.units. Optional 15
profiler.client.period.duration.units The units used to specify the profile period duration. This value should be defined along with profiler.client.period.duration. Optional MINUTES
profiler.client.hbase.table The name of the HBase table used to store profile data. Optional profiler
profiler.client.hbase.column.family The name of the HBase column family used to store profile data. Optional P
profiler.client.salt.divisor The salt divisor used to store profile data. Optional 1000
profiler.default.value The default value to be returned if a profile is not written for a given period for a profile and entity. Optional null
hbase.provider.impl The name of the HBaseTableProvider implementation class. Optional

Profile Selectors

You will notice that the third argument for PROFILE_GET is a list of ProfilePeriod objects. This list is expected to be produced by another Stellar function. There are a couple options available.

PROFILE_FIXED

The profiler periods associated with a fixed lookback starting from now. These are ProfilePeriod objects.

REQUIRED:
    durationAgo - How long ago should values be retrieved from?
    units - The units of 'durationAgo'.
OPTIONAL:
    config_overrides - Optional - Map (in curly braces) of name:value pairs, each overriding the global config parameter
            of the same name. Default is the empty Map, meaning no overrides.

e.g. To retrieve all the profiles for the last 5 hours.  PROFILE_GET('profile', 'entity', PROFILE_FIXED(5, 'HOURS'))

Note that the config_overrides parameter operates exactly as the config_overrides argument in PROFILE_GET. The only available parameters for override are:

  • profiler.client.period.duration
  • profiler.client.period.duration.units

PROFILE_WINDOW

PROFILE_WINDOW is intended to provide a finer-level of control over selecting windows for profiles:

  • Specify windows relative to the data timestamp (see the optional now parameter below)
  • Specify non-contiguous windows to better handle seasonal data (e.g. the last hour for every day for the last month)
  • Specify profile output excluding holidays
  • Specify only profile output on a specific day of the week

It does this by a domain specific language mimicking natural language that defines the windows excluded.

REQUIRED:
    windowSelector - The statement specifying the window to select.
    now - Optional - The timestamp to use for now.
OPTIONAL:
    config_overrides - Optional - Map (in curly braces) of name:value pairs, each overriding the global config parameter
            of the same name. Default is the empty Map, meaning no overrides.

e.g. To retrieve all the measurements written for 'profile' and 'entity' for the last hour 
on the same weekday excluding weekends and US holidays across the last 14 days: 
PROFILE_GET('profile', 'entity', PROFILE_WINDOW('1 hour window every 24 hours starting from 14 days ago including the current day of the week excluding weekends, holidays:us'))

Note that the config_overrides parameter operates exactly as the config_overrides argument in PROFILE_GET. The only available parameters for override are:

  • profiler.client.period.duration
  • profiler.client.period.duration.units
The Profile Selector Language

The domain specific language can be broken into a series of clauses, some optional

One must specify either a total temporal duration or a temporal window width. The remaining clauses are optional. During the course of the following discussion, we will color code the clauses in the examples and link them to the relevant section for more detail.

From a high level, the language fits the following three forms, which are composed of the clauses above:

Total Temporal Duration

Total temporal duration is specified by a phrase: FROM time_interval AGO TO time_interval AGO This indicates the beginning and ending of a time interval. This is an inclusive duration.

  • FROM - Can be the words “from” or “starting from”
  • time_interval - A time amount followed by a unit (e.g. 1 hour). Fractional amounts are not supported. The unit may be “minute”, “day”, “hour” with any pluralization.
  • TO - Can be the words “until” or “to”
  • AGO - Optionally the word “ago”

The TO time_interval AGO portion is optional. If unspecified then it is expected that the time interval ends now.

Due to the vagaries of the english language, the from and the to portions, if both specified, are interchangeable with regard to which one specifies the start and which specifies the end.

In other words “starting from 1 hour ago to 30 minutes ago” and “starting from 30 minutes ago to 1 hour ago” specify the same temporal duration.

Examples

Temporal Window Width

Temporal window width is the specification of a window. A window is may either repeat within total temporal duration or may fill the total temporal duration. This is an inclusive window. A window is specified by the phrase: time_interval WINDOW

  • time_interval - A time amount followed by a unit (e.g. 1 hour). Fractional amounts are not supported. The unit may be “minute”, “day”, “hour” with any pluralization.
  • WINDOW - Optionally the word “window”

Examples

Skip distance

Skip distance is the amount of time between temporal window beginnings that the next window starts. It is, in effect, the window period.

It is specified by the phrase EVERY time_interval

  • time_interval - A time amount followed by a unit (e.g. 1 hour). Fractional amounts are not supported. The unit may be “minute”, “day”, “hour” with any pluralization.
  • EVERY - The word/phrase “every” or “for every”

Examples

Inclusion/Exclusion specifiers

Inclusion and Exclusion specifiers operate as filters on the set of windows. They operate on the window beginning timestamp.

For inclusion specifiers, windows who are passed by any of the set of inclusion specifiers are included.
inclusion specifiers. Similarly, windows who are passed by any of the set of exclusion specifiers are excluded. Exclusion specifiers trump inclusion specifiers.

Specifiers follow one of the following formats depending on if it is an inclusion or exclusion specifier:

  • INCLUSION specifier, specifier, ...
    • INCLUSION can be “include”, “includes” or “including”
  • EXCLUSION specifier, specifier, ...
    • EXCLUSION can be “exclude”, “excludes” or “excluding”

The specifiers are a set of fixed specifiers available as part of the language:

  • Fixed day of week-based specifiers - includes or excludes if the window is on the specified day of the week
    • “monday” or “mondays”
    • “tuesday” or “tuesdays”
    • “wednesday” or “wednesdays”
    • “thursday” or “thursdays”
    • “friday” or “fridays”
    • “saturday” or “saturdays”
    • “sunday” or “sundays”
    • “weekday” or “weekdays”
    • “weekend” or "“weekends”
  • Relative day of week-based specifiers - includes or excludes based on the day of week relative to now
    • “current day of the week”
    • “current day of week”
    • “this day of the week”
    • “this day of week”
  • Specified date - includes or excludes based on the specified date
    • “date” - Takes up to 2 arguments
      • The day in yyyy/MM/dd format if no second argument is provided
      • Optionally the format to specify the first argument in
      • Example: date:2017/12/25 would include or exclude December 25, 2017
      • Example: date:20171225:yyyyMMdd would include or exclude December 25, 2017
  • Holidays - includes or excludes based on if the window starts during a holiday
    • “holiday” or “holidays”
      • Arguments form the jollyday hierarchy of holidays. e.g. “us:nyc” would be holidays for New York City, USA
      • If none is specified, it will choose based on locale.
      • Countries supported are those supported in jollyday
      • Example: holiday:us:nyc would be the holidays of New York City, USA
      • Example: holiday:hu would be the holidays of Hungary

WARNING: Daylight Savings Time effects

While Universal Time (UTC) is nice and constant, many servers are set to local timezones that enable Daylight Savings Time (DST). This means that twice a year, on DST transition weekends, “Sunday” is either 23 or 25 hours long. However, durations specified as “7 days ago” are always interpreted as “7*24 hours ago”. This can lead to some surprising effects when using days of the week as inclusion or exclusion specifiers.

For example, the profile window specified by the phrase “30 minute window every 24 hours from 7 days ago” will always have 7 thirty-minute intervals, and these will normally occur on 5 weekdays and 2 weekend days. However, if you invoke this window at 12:15am any day during the week following the start of DST, you will get these intervals (supposing you start early on a Wednesday morning):

Tuesday 12:15am-12:45am (yesterday)
Monday 12:15am-12:45am
Saturday 11:15pm-11:45pm (skipped Sunday!)
Friday 11:15pm-11:45pm
Thursday 11:15pm-11:45pm
Wednesday 11:15pm-11:45pm
Tuesday 11:15pm-11:45pm

Sunday got skipped over because it was only 23 hours long; that is, there were 24 hours between Saturday 11:15pm and Monday 12:15am. So if you specified “excluding weekends”, you would get 6 days’ intervals instead of the expected 5. There are multiple variations on this theme.

Remember that the underlying time is kept in UTC, so the data is always correct. It is only when attempting to interpret UTC as local time, date, and day, that these confusions may occur. They may be eliminated by setting your server timezone to UTC, or otherwise disabling DST.

Examples

Assume these are executed at noon.

Errors

The most common result of incorrect PROFILE_GET arguments or Client configuration parameters is an empty result set, rather than an error. The Client cannot effectively validate the arguments, because the Profiler configuration parameters may be changed and the profile itself does not store them. The person doing the querying must carry forward the knowledge of the Profiler configuration parameters from the time of profile creation, and use corresponding PROFILE_GET arguments and Client configuration parameters when querying the data.

Examples

The following are usage examples that show how the Stellar API can be used to read profiles generated by the Metron Profiler. This API would be used in conjunction with other Stellar functions like MAAS_MODEL_APPLY to perform model scoring on streaming data.

These examples assume a profile has been defined called ‘snort-alerts’ that tracks the number of Snort alerts associated with an IP address over time. The profile definition might look similar to the following.

{
  "profiles": [
    {
      "profile": "snort-alerts",
      "foreach": "ip_src_addr",
      "onlyif":  "source.type == 'snort'",
      "update":  { "s": "STATS_ADD(s, 1)" },
      "result":  "STATS_MEAN(s)"
    }
  ]
}

During model scoring the entity being scored, in this case a particular IP address, will be known. The following examples shows how this profile data might be retrieved. Retrieve all values of ‘snort-alerts’ from ‘10.0.0.1’ over the past 4 hours.

PROFILE_GET('snort-alerts', '10.0.0.1', PROFILE_FIXED(4, 'HOURS'))

Retrieve all values of ‘snort-alerts’ from ‘10.0.0.1’ over the past 2 days.

PROFILE_GET('snort-alerts', '10.0.0.1', PROFILE_FIXED(2, 'DAYS'))

If the profile had been defined to group the data by weekday versus weekend, then the following example would apply:

Retrieve all values of ‘snort-alerts’ from ‘10.0.0.1’ that occurred on ‘weekdays’ over the past 30 days.

PROFILE_GET('snort-alerts', '10.0.0.1', PROFILE_FIXED(30, 'DAYS'), ['weekdays'] )

The client may need to use a configuration different from the current Client configuration settings. For example, perhaps you are on a cluster shared with other analysts, and need to access a profile that was constructed 2 months ago using different period duration, while they are accessing more recent profiles constructed with the currently configured period duration. For this situation, you may use the config_overrides argument:

Retrieve all values of ‘snort-alerts’ from ‘10.0.0.1’ over the past 2 days, with no groupBy, and overriding the usual global client configuration parameters for window duration.

PROFILE_GET('profile1', 'entity1', PROFILE_FIXED(2, 'DAYS', {'profiler.client.period.duration' : '2', 'profiler.client.period.duration.units' : 'MINUTES'}), [])

Retrieve all values of ‘snort-alerts’ from ‘10.0.0.1’ that occurred on ‘weekdays’ over the past 30 days, overriding the usual global client configuration parameters for window duration.

PROFILE_GET('profile1', 'entity1', PROFILE_FIXED(30, 'DAYS', {'profiler.client.period.duration' : '2', 'profiler.client.period.duration.units' : 'MINUTES'}), ['weekdays'] )

Getting Started

These instructions step through the process of using the Stellar Client API on a live cluster. These instructions assume that the ‘Getting Started’ instructions included with the Metron Profiler have been followed. This will create a Profile called ‘test’ whose data will be retrieved with the Stellar Client API.

To validate that everything is working, login to the server hosting Metron. We will use the Stellar Shell to replicate the execution environment of Stellar running in a Storm topology, like Metron’s Parser or Enrichment topology. Replace ‘node1:2181’ with the URL to a Zookeeper Broker.

[root@node1 0.4.2]# bin/stellar -z node1:2181
Stellar, Go!
Please note that functions are loading lazily in the background and will be unavailable until loaded fully.
{es.clustername=metron, es.ip=node1, es.port=9300, es.date.format=yyyy.MM.dd.HH}

[Stellar]>>> ?PROFILE_GET
Functions loaded, you may refer to functions now...
PROFILE_GET
Description: Retrieves a series of values from a stored profile.

Arguments:
	profile - The name of the profile.
	entity - The name of the entity.
	durationAgo - How long ago should values be retrieved from?
	units - The units of 'durationAgo'.
	groups_list - Optional, must correspond to the 'groupBy' list used in profile creation - List (in square brackets) of 
            groupBy values used to filter the profile. Default is the empty list, meaning groupBy was not used when 
            creating the profile.
	config_overrides - Optional - Map (in curly braces) of name:value pairs, each overriding the global config parameter
            of the same name. Default is the empty Map, meaning no overrides.

Returns: The selected profile measurements.

[Stellar]>>> PROFILE_GET('test','192.168.138.158', 1, 'HOURS')
[12078.0, 8921.0, 12131.0]

The client API call above has retrieved the past hour of the ‘test’ profile for the entity ‘192.168.138.158’.

Developing Profiles

Troubleshooting issues when programming against a live stream of data can be difficult. The Stellar REPL is a powerful tool to help work out the kinds of enrichments and transformations that are needed. The Stellar REPL can also be used to help when developing profiles for the Profiler.

Follow these steps in the Stellar REPL to see how it can be used to help create profiles.

  1. Take a first pass at defining your profile. As an example, in the editor copy/paste the basic “Hello, World” profile below.

    [Stellar]>>> conf := SHELL_EDIT()
    [Stellar]>>> conf
    {
      "profiles": [
        {
          "profile": "hello-world",
          "onlyif":  "exists(ip_src_addr)",
          "foreach": "ip_src_addr",
          "init":    { "count": "0" },
          "update":  { "count": "count + 1" },
          "result":  "count"
        }
      ]
    }
    
  2. Initialize the Profiler.

    [Stellar]>>> profiler := PROFILER_INIT(conf)
    [Stellar]>>> profiler
    Profiler{1 profile(s), 0 messages(s), 0 route(s)}
    

    The profiler itself will show the number of profiles defined, the number of messages applied, and the number of routes taken.

    A route is defined when a message is applied to a specific profile. If a message is applied and not needed by any profile, then there are no routes. If a message is needed by one profile, then one route has been defined. If a message is needed by two profiles, then two routes have been defined.

  3. Create a message to simulate the type of telemetry that you expect to be profiled. As an example, in the editor copy/paste the JSON below.

    [Stellar]>>> message := SHELL_EDIT()
    [Stellar]>>> message
    {
      "ip_src_addr": "10.0.0.1",
      "protocol": "HTTPS",
      "length": "10",
      "bytes_in": "234"
    }
    
  4. Apply some telemetry messages to your profiles. The following applies the same message 3 times.

    [Stellar]>>> PROFILER_APPLY(message, profiler)
    Profiler{1 profile(s), 1 messages(s), 1 route(s)}
    
    [Stellar]>>> PROFILER_APPLY(message, profiler)
    Profiler{1 profile(s), 2 messages(s), 2 route(s)}
    
    [Stellar]>>> PROFILER_APPLY(message, profiler)
    Profiler{1 profile(s), 3 messages(s), 3 route(s)}
    

    It is also possible to apply multiple messages at once. This is useful when testing against a larger set of data. To do this, create a string that contains a JSON array of messages and pass that to the PROFILER_APPLY function.

  5. Flush the Profiler to see what has been calculated. A flush is what occurs at the end of each 15 minute period in the Profiler. The result is a list of profile measurements. Each measurement is a map containing detailed information about the profile data that has been generated.

    [Stellar]>>> values := PROFILER_FLUSH(profiler)
    [Stellar]>>> values
    [{period={duration=900000, period=1669628, start=1502665200000, end=1502666100000}, 
       profile=hello-world, groups=[], value=3, entity=10.0.0.1}]
    

    This profile simply counts the number of messages by IP source address. Notice that the value is ‘3’ for the entity ‘10.0.0.1’ as we applied 3 messages with an ‘ip_src_addr’ of ‘10.0.0.1’. There will always be one measurement for each [profile, entity] pair.

  6. If you are unhappy with the data that has been generated, then ‘wash, rinse and repeat’ this process. Once you are happy with the profile that was created, follow the Getting Started guide to use the profile against your live, streaming data in a Metron cluster.